evaluation.html

<h2>Submission and Evaluation</h2>

</br>
<p> The challenge is only evaluated with automatic metrics. We will use BLEU, TER, chrF (sacreBLEU versions) and COMET. chrF is the official evaluation metric. COMET might be misleading as it does not include Wayúu (but it does Spanish). When submitting a run, be patient, COMET runs on CPU. It might take 20 minutes before your results appear in the leaderboard.</p>

</br>
<p><strong>Phase I, development</strong>: Name your file tatoeba.test.guc2esp.esp and zip it. Upload your zipped translation. Notice that only zip compression works, and non-zipped files will result in an error.</p>

<p><strong>Phase II, evaluation</strong>:  Name your file general.test.guc2esp.esp and zip it. Upload your zipped translation. Notice that only zip compression works, and non-zipped files will result in an error.</p>